#!/usr/bin/env python
import sys, os.path, datetime, codecs
import sqlite3
SCRIPT_DIR = os.path.dirname(os.path.realpath(__file__))
# All text is saved to OUT_FILE
OUT_FILE = os.path.join(SCRIPT_DIR, "..", "srcblog", "evernote-utf8.txt")
EVERNOTE_DIR = os.path.join(os.path.expanduser("~"), "Library", "Application Support", "Evernote", "data")
# this might be different, depending on the Evernote version
EVERNOTE_DIR = os.path.join(EVERNOTE_DIR, "40145")
EVERNOTE_DB_PATH = os.path.join(EVERNOTE_DIR, "Evernote.sql")
EVERNOTE_CONTENT_DIR = os.path.join(EVERNOTE_DIR, "content")
g_conn = None
def get_conn():
global g_conn
if not g_conn:
g_conn = sqlite3.connect(EVERNOTE_DB_PATH)
return g_conn
def close_conn():
global g_conn
if g_conn:
g_conn.close()
g_conn = None
g_article_to_tags_map = None
def get_article_to_tags_map():
global g_article_to_tags_map
if g_article_to_tags_map:
return g_article_to_tags_map
conn = get_conn()
c = conn.cursor()
c.execute("select Z_3NOTES, Z_7TAGS from Z_3TAGS")
d = {}
for row in c:
article_id = row[0]
tag_id = row[1]
if article_id in d:
d[article_id].append(tag_id)
else:
d[article_id] = [tag_id]
c.close()
g_article_to_tags_map = d
return g_article_to_tags_map
g_tag_to_name_map = None
def get_tag_to_name_map():
global g_tag_to_name_map
if g_tag_to_name_map:
return g_tag_to_name_map
conn = get_conn()
c = conn.cursor()
c.execute("select Z_PK, ZNAME2 from ZENSERVICEENTITY where Z_ENT = 7")
d = {}
for row in c:
tag_id = row[0]
tag_name = row[1]
d[tag_id] = tag_name
c.close()
g_tag_to_name_map = d
return g_tag_to_name_map
def tags_for_article(article_id):
article_to_tags_map = get_article_to_tags_map()
if article_id not in article_to_tags_map:
return None
tag_ids = article_to_tags_map[article_id]
tag_to_name_map = get_tag_to_name_map()
tags = [tag_to_name_map[tag_id] for tag_id in tag_ids]
return tags
def datetime_to_str(dt): return dt.strftime('%Y-%m-%d %H:%M:%S')
def get_content(content_enml_path):
fo = codecs.open(content_enml_path, "r", "utf-8")
txt = fo.read()
fo.close()
en_note_tag_start = txt.find("<en-note")
assert -1 != en_note_tag_start
en_note_tag_end = txt.find(">", en_note_tag_start)
assert -1 != en_note_tag_end
en_note_end = txt.find("</en-note", en_note_tag_end)
assert -1 != en_note_end
txt = txt[en_note_tag_end+1:en_note_end]
return txt.encode("utf-8")
def extract_articles():
txt_lines = []
conn = get_conn()
c = conn.cursor()
c.execute("select Z_PK, ZTITLE, ZCREATED from ZENSERVICEENTITY where Z_ENT = 3")
# ZCREATED timestamp is in weird format that looks like 31 years after
# unix timestamp. so this is a crude way to approximate this. Might be off
# by a day or so
td = datetime.timedelta(days=31*365+9)
for row in c:
article_id = row[0]
title = row[1]
#created_on = datetime.datetime.fromtimestamp(row[2])
created_on = datetime.datetime.fromtimestamp(row[2])
created_on = created_on + td
title_line = u"@Title: %s" % title
print(title_line)
date_line = "@Date: %s" % datetime_to_str(created_on)
print(date_line)
tags = tags_for_article(article_id)
tags_line_utf8 = None
if tags:
tags_line = "@Tags: %s" % u", ".join(tags)
tags_line_utf8 = tags_line.encode("utf-8")
print(tags_line)
content_enml_path = os.path.join(EVERNOTE_CONTENT_DIR, "p%s" % article_id, "content.enml")
content_utf8 = get_content(content_enml_path).strip()
if 0 == len(content_utf8):
print("\n!!Skipping '%s'\n" % title)
continue
#print(content_utf8)
#print("")
txt_lines.append(title_line.encode("utf-8"))
txt_lines.append(date_line.encode("utf-8"))
if tags_line_utf8:
txt_lines.append(tags_line_utf8)
txt_lines.append(content_utf8)
txt_lines.append("")
txt_utf8 = "\n".join(txt_lines)
open(OUT_FILE, "w").write(txt_utf8)
def main():
extract_articles()
close_conn()
if __name__ == "__main__":
main()
