Merge pull request #50 from kidd/add-firefox-history

Add firefox history module to memacs

bin/memacs_firefox.py

@@ -0,0 +1,33 @@
+#!/usr/bin/env python2
+# -*- coding: utf-8 -*-
+# Time-stamp: <2013-04-04 16:18:07 vk>
+
+from memacs.firefox import Firefox
+
+PROG_VERSION_NUMBER = u"0.0"
+PROG_VERSION_DATE = u"2018-07-14"
+PROG_SHORT_DESCRIPTION = u"Memacs for firefox url history "
+PROG_TAG = u"firefox"
+PROG_DESCRIPTION = u"""
+This class will parse firefox history file (places.sqlite) and
+produce an org file with all your visited sites
+"""
+# set CONFIG_PARSER_NAME only, when you want to have a config file
+# otherwise you can comment it out
+# CONFIG_PARSER_NAME="memacs-example"
+COPYRIGHT_YEAR = "2018"
+COPYRIGHT_AUTHORS = """Raimon Grau <[email protected]>"""
+
+
+if __name__ == "__main__":
+    memacs = Firefox(
+        prog_version=PROG_VERSION_NUMBER,
+        prog_version_date=PROG_VERSION_DATE,
+        prog_description=PROG_DESCRIPTION,
+        prog_short_description=PROG_SHORT_DESCRIPTION,
+        prog_tag=PROG_TAG,
+        copyright_year=COPYRIGHT_YEAR,
+        copyright_authors=COPYRIGHT_AUTHORS
+#       use_config_parser_name=CONFIG_PARSER_NAME
+        )
+    memacs.handle_main()

docs/memacs_firefox_history.org

@@ -0,0 +1,31 @@
+## Time-stamp: <2018-07-21>
+## This file is best viewed with GNU Emacs Org-mode: http://orgmode.org/
+
+* memacs-firefox_history
+** Options
+
+   - ~-f~, ~--file~, path to a sqlite db file which contains the firefox history: ~~/.mozilla/firefox/123adsf1.default/places.sqlite~.
+   - ~--output-format~, format for the headers of the generated org. available tags are: ~guid, url, title, visit_count, timestamp~
+
+** Data Source
+
+The visited sites in Firefox are stored in a ~places.sqlite~ file that
+contains every url you visited along with the last visited date.
+
+We can use Memacs_firefox to generate a file with the datetimes of the urls:
+
+** Example Invocation
+*** Example with rev-list file:
+: python bin/memacs_firefox.py -f "/home/rgrau/.mozilla/firefox/3hlzuwen.default/places.sqlite" >~/org/mozhist.org_archive
+
+** Example Orgmode entries
+
+   With the default --output-format, a regular entry looks like this:
+
+: * Memacs for firefox url history           :Memacs:firefox:
+: ** <2018-07-20 Fri 23:51> [[https://github.com/kidd][kidd (Raimon Grau (rgrau))]]
+:   :PROPERTIES:
+:   :URL:         https://github.com/kidd
+:   :VISIT_COUNT: 1
+:   :ID:          68ff284b25379144784a214e7a0d19631c3d18f1
+:   :END:

memacs/firefox.py

@@ -0,0 +1,85 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# Time-stamp: <2017-02-07 19:25 manu>
+
+import sqlite3
+import datetime
+import sys
+import os
+import re
+
+from lib.orgproperty import OrgProperties
+from lib.orgformat import OrgFormat
+from lib.memacs import Memacs
+
+reload(sys)
+sys.setdefaultencoding('utf8')
+
+class Firefox(Memacs):
+    def _parser_add_arguments(self):
+        """
+        overwritten method of class Memacs
+
+        add additional arguments
+        """
+        Memacs._parser_add_arguments(self)
+
+        self._parser.add_argument(
+            "-f", "--file", dest="historystore",
+            action="store", type=file, required=True,
+            help="""path to places.sqlite file. usually in
+/home/rgrau/.mozilla/firefox/__SOMETHING__.default/places.sqlite """)
+
+        self._parser.add_argument(
+            "--output-format", dest="output_format",
+            action="store", default="[[{url}][{title}]]",
+            help="format string to use for the headline")
+
+
+    def _parser_parse_args(self):
+        """
+        overwritten method of class Memacs
+
+        all additional arguments are parsed in here
+        """
+        Memacs._parser_parse_args(self)
+
+    def _handle_url(self, params):
+        timestamp = datetime.datetime.fromtimestamp(int(params['timestamp']/1000000))
+
+        properties = OrgProperties()
+        if (params['title'] == "") :
+            params['title'] = params['url']
+        properties.add('URL', params['url'])
+        properties.add('VISIT_COUNT', params['visit_count'])
+
+        output = ""
+        try:
+            output = self._args.output_format.decode('utf-8').format(**params)
+        except Exception:
+            pass
+
+        self._writer.write_org_subitem(
+            timestamp=OrgFormat.datetime(timestamp),
+            output=output, properties=properties)
+
+    def _main(self):
+        """
+        get's automatically called from Memacs class
+        """
+        conn = sqlite3.connect(os.path.abspath(self._args.historystore.name))
+        query = conn.execute("""
+        select url, title, visit_count,
+               -- datetime(last_visit_date/1000000, 'unixepoch')
+                last_visit_date
+        from   moz_places
+        where  last_visit_date IS NOT NULL
+        order by last_visit_date """)
+
+        for row in query:
+            self._handle_url({
+                'url'         : row[0],
+                'title'       : row[1],
+                'visit_count' : row[2],
+                'timestamp'   : row[3],
+            })