From 27e5342e822900d11224bc89cf4e3df51b1108be Mon Sep 17 00:00:00 2001
From: Markus Birth <mbirth@gmail.com>
Date: Wed, 5 Jan 2022 14:04:46 +0100
Subject: [PATCH] Initial commit

---
 .gitignore                    |  4 ++
 Pipfile                       | 11 +++++
 Pipfile.lock                  | 70 +++++++++++++++++++++++++++++++
 README.md                     | 38 +++++++++++++++++
 config.yaml.example           | 33 +++++++++++++++
 jekyll2s9y.py                 | 78 +++++++++++++++++++++++++++++++++++
 jekyllreader/__init__.py      |  2 +
 jekyllreader/jekyllarticle.py | 69 +++++++++++++++++++++++++++++++
 jekyllreader/jekyllreader.py  | 25 +++++++++++
 s9ywriter/__init__.py         |  2 +
 s9ywriter/s9yentry.py         | 14 +++++++
 s9ywriter/s9ywriter.py        | 75 +++++++++++++++++++++++++++++++++
 12 files changed, 421 insertions(+)
 create mode 100644 .gitignore
 create mode 100644 Pipfile
 create mode 100644 Pipfile.lock
 create mode 100644 README.md
 create mode 100644 config.yaml.example
 create mode 100755 jekyll2s9y.py
 create mode 100644 jekyllreader/__init__.py
 create mode 100644 jekyllreader/jekyllarticle.py
 create mode 100644 jekyllreader/jekyllreader.py
 create mode 100644 s9ywriter/__init__.py
 create mode 100644 s9ywriter/s9yentry.py
 create mode 100644 s9ywriter/s9ywriter.py

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..4cbc351
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,4 @@
+*.db
+jekyll/
+uploads/
+/config.yaml
diff --git a/Pipfile b/Pipfile
new file mode 100644
index 0000000..005d615
--- /dev/null
+++ b/Pipfile
@@ -0,0 +1,11 @@
+[[source]]
+url = "https://pypi.python.org/simple"
+verify_ssl = true
+name = "pypi"
+
+[packages]
+python-frontmatter = "*"
+pytz = "*"
+pyyaml = "*"
+
+[dev-packages]
diff --git a/Pipfile.lock b/Pipfile.lock
new file mode 100644
index 0000000..902a1d1
--- /dev/null
+++ b/Pipfile.lock
@@ -0,0 +1,70 @@
+{
+    "_meta": {
+        "hash": {
+            "sha256": "80e20d60ab55b386ef8a6294cfd3e9712ae0ea8aeb23582f66e8534f61361e74"
+        },
+        "pipfile-spec": 6,
+        "requires": {},
+        "sources": [
+            {
+                "name": "pypi",
+                "url": "https://pypi.python.org/simple",
+                "verify_ssl": true
+            }
+        ]
+    },
+    "default": {
+        "python-frontmatter": {
+            "hashes": [
+                "sha256:766ae75f1b301ffc5fe3494339147e0fd80bc3deff3d7590a93991978b579b08",
+                "sha256:e98152e977225ddafea6f01f40b4b0f1de175766322004c826ca99842d19a7cd"
+            ],
+            "index": "pypi",
+            "version": "==1.0.0"
+        },
+        "pytz": {
+            "hashes": [
+                "sha256:83a4a90894bf38e243cf052c8b58f381bfe9a7a483f6a9cab140bc7f702ac4da",
+                "sha256:eb10ce3e7736052ed3623d49975ce333bcd712c7bb19a58b9e2089d4057d0798"
+            ],
+            "index": "pypi",
+            "version": "==2021.1"
+        },
+        "pyyaml": {
+            "hashes": [
+                "sha256:08682f6b72c722394747bddaf0aa62277e02557c0fd1c42cb853016a38f8dedf",
+                "sha256:0f5f5786c0e09baddcd8b4b45f20a7b5d61a7e7e99846e3c799b05c7c53fa696",
+                "sha256:129def1b7c1bf22faffd67b8f3724645203b79d8f4cc81f674654d9902cb4393",
+                "sha256:294db365efa064d00b8d1ef65d8ea2c3426ac366c0c4368d930bf1c5fb497f77",
+                "sha256:3b2b1824fe7112845700f815ff6a489360226a5609b96ec2190a45e62a9fc922",
+                "sha256:3bd0e463264cf257d1ffd2e40223b197271046d09dadf73a0fe82b9c1fc385a5",
+                "sha256:4465124ef1b18d9ace298060f4eccc64b0850899ac4ac53294547536533800c8",
+                "sha256:49d4cdd9065b9b6e206d0595fee27a96b5dd22618e7520c33204a4a3239d5b10",
+                "sha256:4e0583d24c881e14342eaf4ec5fbc97f934b999a6828693a99157fde912540cc",
+                "sha256:5accb17103e43963b80e6f837831f38d314a0495500067cb25afab2e8d7a4018",
+                "sha256:607774cbba28732bfa802b54baa7484215f530991055bb562efbed5b2f20a45e",
+                "sha256:6c78645d400265a062508ae399b60b8c167bf003db364ecb26dcab2bda048253",
+                "sha256:72a01f726a9c7851ca9bfad6fd09ca4e090a023c00945ea05ba1638c09dc3347",
+                "sha256:74c1485f7707cf707a7aef42ef6322b8f97921bd89be2ab6317fd782c2d53183",
+                "sha256:895f61ef02e8fed38159bb70f7e100e00f471eae2bc838cd0f4ebb21e28f8541",
+                "sha256:8c1be557ee92a20f184922c7b6424e8ab6691788e6d86137c5d93c1a6ec1b8fb",
+                "sha256:bb4191dfc9306777bc594117aee052446b3fa88737cd13b7188d0e7aa8162185",
+                "sha256:bfb51918d4ff3d77c1c856a9699f8492c612cde32fd3bcd344af9be34999bfdc",
+                "sha256:c20cfa2d49991c8b4147af39859b167664f2ad4561704ee74c1de03318e898db",
+                "sha256:cb333c16912324fd5f769fff6bc5de372e9e7a202247b48870bc251ed40239aa",
+                "sha256:d2d9808ea7b4af864f35ea216be506ecec180628aced0704e34aca0b040ffe46",
+                "sha256:d483ad4e639292c90170eb6f7783ad19490e7a8defb3e46f97dfe4bacae89122",
+                "sha256:dd5de0646207f053eb0d6c74ae45ba98c3395a571a2891858e87df7c9b9bd51b",
+                "sha256:e1d4970ea66be07ae37a3c2e48b5ec63f7ba6804bdddfdbd3cfd954d25a82e63",
+                "sha256:e4fac90784481d221a8e4b1162afa7c47ed953be40d31ab4629ae917510051df",
+                "sha256:fa5ae20527d8e831e8230cbffd9f8fe952815b2b7dae6ffec25318803a7528fc",
+                "sha256:fd7f6999a8070df521b6384004ef42833b9bd62cfee11a09bda1079b4b704247",
+                "sha256:fdc842473cd33f45ff6bce46aea678a54e3d21f1b61a7750ce3c498eedfe25d6",
+                "sha256:fe69978f3f768926cfa37b867e3843918e012cf83f680806599ddce33c2c68b0"
+            ],
+            "index": "pypi",
+            "version": "==5.4.1"
+        }
+    },
+    "develop": {}
+}
diff --git a/README.md b/README.md
new file mode 100644
index 0000000..d61b428
--- /dev/null
+++ b/README.md
@@ -0,0 +1,38 @@
+jekyll2s9y importer
+===================
+
+This is a tool to import my Jekyll-based [GitHub Pages](https://pages.github.com) blog/wiki
+into [Serendipity](https://s9y.org).
+
+
+Preparations
+------------
+
+Setup Serendipity with an SQLite database. Download the database file to where you also plan to run jekyll2s9y.
+
+
+Usage
+-----
+
+Copy `config.yaml.example` to `config.yaml` and modify according to your needs. Note that
+`jekyll_dir` has to point to your Jekyll base directory.
+
+If not done already, update the Python environment:
+
+    pipenv install
+
+Then run the script:
+
+    pipenv run ./jekyll2s9y.py
+
+Now copy the new database file (`s9y_database_output` in the config.yaml) back to your server and overwrite
+the old version. Also copy the `uploads` directory containing the media files. Then, in the Serendipity admin
+area, edit your configuration and change the permalink for "Entry URL structure" in any way (e.g. add a letter
+to the end) to have the permalinks regenerated. Afterwards, you can undo the change again.
+
+
+Issues
+------
+
+If there are two entries where the permalink generation (i.e. stripping all emojis, etc.) results in the same
+permalink, only the older one can be accessed.
diff --git a/config.yaml.example b/config.yaml.example
new file mode 100644
index 0000000..9ae6d5f
--- /dev/null
+++ b/config.yaml.example
@@ -0,0 +1,33 @@
+general:
+  jekyll_dir: jekyll
+  s9y_database: serendipity.db
+  s9y_media_dir: uploads
+  s9y_database_output: serendipity_new.db
+
+  # Timezone for stamps without one
+  timezone: Europe/Berlin
+
+# As Jekyll only knows tags, specify which tags are to be
+# considered s9y categories. Tags not mentioned here will
+# be kept as tags (serendipity_plugin_freetag)
+categories:
+  - know-how
+  - development
+  - gtd
+  - review
+  - hacking
+  - hardware
+  - software
+  - miscellaneous
+
+s9y_defaults:
+  comments: 0
+  trackbacks: 0
+  exflag: 1
+  # CHANGE THIS TO YOUR Serendipity Username
+  author: mbirth
+  # CHANGE THIS TO YOUR Serendipity User-ID
+  authorid: 1
+  isdraft: "false"
+  allow_comments: "true"
+  moderate_comments: "false"
diff --git a/jekyll2s9y.py b/jekyll2s9y.py
new file mode 100755
index 0000000..bc5d366
--- /dev/null
+++ b/jekyll2s9y.py
@@ -0,0 +1,78 @@
+#!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+from datetime import datetime
+from os import makedirs
+from os.path import basename
+import shutil
+import pytz
+import yaml
+import jekyllreader
+import s9ywriter
+
+
+with open("config.yaml", "rt") as f:
+    config = yaml.load(f)
+
+print(repr(config))
+
+# copy original file to working copy
+shutil.copyfile(config["general"]["s9y_database"], config["general"]["s9y_database_output"])
+
+DATEFORMAT_IN = "%Y-%m-%d %H:%M:%S %z"
+DATEFORMAT_OUT = "%Y-%m-%d %H:%M:%S"
+
+LOCAL_TIMEZONE = pytz.timezone(config["general"]["timezone"])
+
+# MAIN SCRIPT
+jk = jekyllreader.JekyllReader(config["general"]["jekyll_dir"])
+s9y = s9ywriter.S9YWriter(config["general"]["s9y_database_output"])
+for i in range(0, jk.len()):
+    print(f"Item: {i}")
+    jk_article = jk.get(i)
+
+    new_entry = s9ywriter.S9YEntry()
+    for k, v in config["s9y_defaults"].items():
+        setattr(new_entry, k, v)
+
+    new_entry.title = jk_article.metadata["title"]
+    date_created = datetime.strptime(jk_article.metadata["created"], DATEFORMAT_IN).astimezone(LOCAL_TIMEZONE)
+    date_updated = datetime.strptime(jk_article.metadata["updated"], DATEFORMAT_IN).astimezone(LOCAL_TIMEZONE)
+    new_entry.timestamp = int(date_created.timestamp())
+    new_entry.last_modified = int(date_updated.timestamp())
+
+    # Handle images
+    img_target_dir = config["general"]["s9y_media_dir"] + "/" + str(date_created.year) + "/"
+    img_files = jk_article.images
+    print(repr(img_files))
+    jk_article.replace_imagepaths("/" + img_target_dir)
+    for img in img_files:
+        img_name = basename(img)
+        makedirs(img_target_dir, exist_ok=True)
+        shutil.copyfile(img, img_target_dir + img_name)
+
+    # Handle body: Split into body+extended if possible
+    content = jk_article.body
+    content = content.replace("\r", "")
+    splits = content.split("\n\n", 1)
+    if len(splits) == 1:
+        new_entry.body = content
+    else:
+        splits[1] = splits[1].strip("\n")
+        (new_entry.body, new_entry.extended) = splits
+
+    # Handle tags/categories and other metadata
+    for t in jk_article.metadata["tags"]:
+        if t in config["categories"]:
+            new_entry.categories.append(t)
+        else:
+            new_entry.tags.append(t)
+
+    if "language" in jk_article.metadata and jk_article.metadata["language"] != "en":
+        if jk_article.metadata["language"] == "de":
+            new_entry.title += " 🇩🇪"
+        else:
+            new_entry.title += " (" + jk_article.metadata["language"] + ")"
+
+    s9y.add_entry(new_entry)
+s9y.commit()
diff --git a/jekyllreader/__init__.py b/jekyllreader/__init__.py
new file mode 100644
index 0000000..f37d1bd
--- /dev/null
+++ b/jekyllreader/__init__.py
@@ -0,0 +1,2 @@
+from .jekyllreader import JekyllReader
+from .jekyllarticle import JekyllArticle
diff --git a/jekyllreader/jekyllarticle.py b/jekyllreader/jekyllarticle.py
new file mode 100644
index 0000000..8610374
--- /dev/null
+++ b/jekyllreader/jekyllarticle.py
@@ -0,0 +1,69 @@
+from functools import partial
+from os.path import basename
+import re
+import frontmatter
+
+
+class JekyllArticle():
+    RE_IMG = r'!\[(.*?)\]\((.+?)( [\'"].+[\'"])?\)'
+    RE_HTML_IMG = r'<img src="(.*?)"'
+    RE_SITE_URL = r'\{\{ ?site\.url ?\}\}'
+
+    def __init__(self, md_file: str, base_dir: str = "."):
+        self.file_path = md_file
+        self.base_dir = base_dir
+        self.body = ""
+        self.metadata = {}
+        self.images = []
+        self.replace_image_newpath = ""
+        self.parse_file()
+        self.collect_images()
+
+    def parse_file(self):
+        article = frontmatter.load(self.file_path)
+        self.metadata = article.metadata
+        self.body = article.content
+
+        # Convert Twig highlights to Markdown Extra highlights
+        self.body = re.sub(r'\{% highlight( (\S+)) %\}', r'```\2', self.body)
+        self.body = re.sub(r'\{% endhighlight %\}', r'```', self.body)
+
+    def collect_images(self):
+        matches = re.findall(self.RE_IMG, self.body)   # Returns a tuple
+        for match in matches:
+            imgfile = match[1]
+            imgfile = re.sub(self.RE_SITE_URL, self.base_dir, imgfile)
+            self.images.append(imgfile)
+        matches = re.findall(self.RE_HTML_IMG, self.body)   # Returns plain matches
+        for match in matches:
+            imgfile = match
+            imgfile = re.sub(self.RE_SITE_URL, self.base_dir, imgfile)
+            self.images.append(imgfile)
+        self.images = list(set(self.images))
+
+    def _replace_single_imagepath(self, match, new_base_url: str = ""):
+        old_imgfile = match.group(2)
+        old_imgfile = re.sub(self.RE_SITE_URL, self.base_dir, old_imgfile)
+        img_name = basename(old_imgfile)
+        new_imgfile = "{}{}".format(new_base_url, img_name)
+        # print("Copying image {} to {} ...".format(old_imgfile, new_imgfile))
+        # shutil.copyfile(old_imgfile, new_imgfile)
+        img_title = ""
+        if match.group(3):
+            img_title = match.group(3)
+        new_string = "![{}]({}{})".format(match.group(1), new_imgfile, img_title)
+        return new_string
+
+    def _replace_single_htmlimagepath(self, match, new_base_url: str = ""):
+        old_imgfile = match.group(1)
+        old_imgfile = re.sub(self.RE_SITE_URL, self.base_dir, old_imgfile)
+        img_name = basename(old_imgfile)
+        new_imgfile = "{}{}".format(new_base_url, img_name)
+        new_string = "<img src=\"" + new_imgfile + "\""
+        return new_string
+
+    def replace_imagepaths(self, new_base_url: str = ""):
+        # Images OLD: ![alt text]({{ site.url }}/assets/blah.jpg "title")
+        # Images NEW: ![alt text](blah.jpg "title")
+        self.body = re.sub(self.RE_IMG, partial(self._replace_single_imagepath, new_base_url=new_base_url), self.body)
+        self.body = re.sub(self.RE_HTML_IMG, partial(self._replace_single_htmlimagepath, new_base_url=new_base_url), self.body)
diff --git a/jekyllreader/jekyllreader.py b/jekyllreader/jekyllreader.py
new file mode 100644
index 0000000..6c76ae8
--- /dev/null
+++ b/jekyllreader/jekyllreader.py
@@ -0,0 +1,25 @@
+import os
+from .jekyllarticle import JekyllArticle
+
+class JekyllReader():
+
+    def __init__(self, src_dir: str):
+        self.src_dir = src_dir
+        self.file_list = []
+        self.find_articles()
+
+    def find_articles(self):
+        for root, dirs, files in os.walk(self.src_dir):
+            if root.split("/")[-1] in ["assets", "css", "images", "fonts", "javascripts", "_includes", "_layouts", self.src_dir]:
+                continue
+            for f in files:
+                if f.split(".")[-1] != "md":
+                    continue
+                filepath = "{}/{}".format(root, f)
+                self.file_list.append(filepath)
+
+    def len(self):
+        return len(self.file_list)
+
+    def get(self, idx: int):
+        return JekyllArticle(self.file_list[idx], self.src_dir)
diff --git a/s9ywriter/__init__.py b/s9ywriter/__init__.py
new file mode 100644
index 0000000..a60f924
--- /dev/null
+++ b/s9ywriter/__init__.py
@@ -0,0 +1,2 @@
+from .s9ywriter import S9YWriter
+from .s9yentry import S9YEntry
diff --git a/s9ywriter/s9yentry.py b/s9ywriter/s9yentry.py
new file mode 100644
index 0000000..18ac6f1
--- /dev/null
+++ b/s9ywriter/s9yentry.py
@@ -0,0 +1,14 @@
+class S9YEntry():
+    def __init__(self):
+        self.db_keys = ["id", "title", "timestamp", "body", "comments", "trackbacks", "extended", "exflag", "author", "authorid", "isdraft", "allow_comments", "last_modified", "moderate_comments"]
+        self.categories = []
+        self.tags = []
+
+    def get_db_insert_values(self):
+        values = []
+        for k in self.db_keys:
+            if hasattr(self, k):
+                values.append(getattr(self, k))
+            else:
+                values.append(None)
+        return values
diff --git a/s9ywriter/s9ywriter.py b/s9ywriter/s9ywriter.py
new file mode 100644
index 0000000..92d5bd3
--- /dev/null
+++ b/s9ywriter/s9ywriter.py
@@ -0,0 +1,75 @@
+import sqlite3
+from .s9yentry import S9YEntry
+
+
+class S9YWriter():
+    def __init__(self, db_file: str):
+        self.db_file = db_file
+        self.db = sqlite3.connect(self.db_file)
+
+    def __del__(self):
+        self.db.close()
+
+    def add_entry(self, entry: S9YEntry):
+        keys = entry.db_keys[1:]
+        placeholders = ["?"] * len(keys)
+        data = entry.get_db_insert_values()[1:]
+        sql = "INSERT INTO entries (" + ", ".join(keys) + ") VALUES (" + ", ".join(placeholders) + ")"
+        #print(sql)
+        result = self.db.execute(sql, data)
+        entry_id = result.lastrowid
+        print(repr(result.lastrowid))
+        print(entry.title)
+
+        # Add permission
+        sql = "INSERT INTO entryproperties VALUES (?, ?, ?)"
+        self.db.execute(sql, [entry_id, "ep_access", "public"])
+
+        # TODO: Add permalink
+        # permalinks (permalink, entry_id, type, data)
+        # permalink = archives/YYYY-MM-DD-%title%.html
+        # %title% = sanitised, no Unicode, ü --> ue, etc.
+        # WORKAROUND: Edit your blog settings and change the permalink to let S9Y regenerate them
+        #             then change back to desired value
+
+        for category in list(set(entry.categories)):
+            self.add_category(entry_id, category)
+
+        for tag in list(set(entry.tags)):
+            self.add_tag(entry_id, tag)
+
+    def commit(self):
+        self.db.commit()
+
+    def add_category(self, entry_id: int, category_name: str):
+        # Tables: category, entrycat, access
+        # Category: categoryid, category_name, "", "", 0, 0, 0, parentid, NULL, NULL
+        # entrycat: entryid, categoryid
+        print(f"{entry_id} - {category_name}")
+        sql = "SELECT categoryid FROM category WHERE category_name = ?"
+        result = self.db.execute(sql, [category_name])
+        cat = result.fetchone()
+        if cat:
+            category_id = cat[0]
+        else:
+            # Category does not yet exist, add it
+            sql = "INSERT INTO category (category_name) VALUES (?)"
+            result = self.db.execute(sql, [category_name])
+            category_id = result.lastrowid
+            # Add access permissions
+            # access: 0, category_id, "category", read
+            # access: 0, category_id, "category", write
+            sql = "INSERT INTO access VALUES (?, ?, ?, ?, ?)"
+            self.db.execute(sql, [0, category_id, "category", "read", ""])
+            self.db.execute(sql, [0, category_id, "category", "write", ""])
+            # Add permalink
+            sql = "INSERT INTO permalinks VALUES (?, ?, ?, ?)"
+            self.db.execute(sql, [f"categories/{category_name}", category_id, "category", None])
+
+        sql = "INSERT INTO entrycat VALUES (?, ?)"
+        self.db.execute(sql, [entry_id, category_id])
+
+    def add_tag(self, entry_id: int, tag: str):
+        # Table: entrytags (entryid, tag)
+        sql = "INSERT INTO entrytags VALUES (?, ?)"
+        self.db.execute(sql, (entry_id, tag))