From bb219e150009b2aa4c0c38439277ba22ab3078f4 Mon Sep 17 00:00:00 2001
From: Markus Birth <mbirth@gmail.com>
Date: Sun, 19 Dec 2021 21:59:14 +0100
Subject: [PATCH] Added image file processing and made syntax check happy.

---
 jekyll2grav.py | 75 +++++++++++++++++++++++++++++++++-----------------
 1 file changed, 49 insertions(+), 26 deletions(-)

diff --git a/jekyll2grav.py b/jekyll2grav.py
index 38e9eb5..b23609a 100755
--- a/jekyll2grav.py
+++ b/jekyll2grav.py
@@ -3,6 +3,7 @@
 
 from datetime import datetime
 import os
+import shutil
 import frontmatter
 import pytz
 import re
@@ -13,28 +14,41 @@ with open("config.yaml", "rt") as f:
 
 print(repr(config))
 
-SRCDIR=config["general"]["jekyll_dir"]
-GRAVDIR=config["general"]["grav_dir"]
+SRCDIR = config["general"]["jekyll_dir"]
+GRAVDIR = config["general"]["grav_dir"]
 
 GRAV_TYPE = config["grav_defaults"]["item_type"]
 
 # Translates Jekyll top level folders to GRAV (sorted)
-FIRST_LEVEL = {
-    "know-how": "03.know-how",
-    "misc": "04.misc",
-    "reviews": "05.reviews",
-    "software": "06.software"
-}
+FIRST_LEVEL = config["jekyll2grav_directories"]
 
-DATEFORMAT_IN="%Y-%m-%d %H:%M:%S %z"
-DATEFORMAT_OUT="%Y-%m-%d %H:%M:%S"
+DATEFORMAT_IN = "%Y-%m-%d %H:%M:%S %z"
+DATEFORMAT_OUT = "%Y-%m-%d %H:%M:%S"
 
-LOCAL_TIMEZONE=pytz.timezone(config["general"]["timezone"])
+LOCAL_TIMEZONE = pytz.timezone(config["general"]["timezone"])
 
 # Jekyll-tags that are considered to be categories, not tags
-CATEGORIES=["know-how", "development", "review", "hacking", "hardware", "software", "miscellaneous"]
+CATEGORIES = config["grav_categories"]
+
+
+def replace_image(match):
+    global target_path
+    old_imgfile = match.group(2)
+    old_imgfile = re.sub(r'\{\{ ?site\.url ?\}\}', SRCDIR, old_imgfile)
+    img_name = os.path.basename(old_imgfile)
+    new_imgfile = "{}/{}".format(target_path, 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), img_name, img_title)
+    return new_string
+
 
 def convert_file(filepath):
+    global target_path
+
     print("Loading {}...".format(filepath))
     post = frontmatter.load(filepath)
 
@@ -50,10 +64,9 @@ def convert_file(filepath):
     post["date"] = date_created.strftime(DATEFORMAT_OUT)
     post["modified_date"] = date_created.strftime(DATEFORMAT_OUT)
 
-
     post["taxonomy"] = {
-        "category": config["grav_defaults"]["categories"],
-        "tag": config["grav_defaults"]["tags"],
+        "category": list(config["grav_defaults"]["categories"]),
+        "tag": list(config["grav_defaults"]["tags"]),
     }
 
     for t in post["tags"]:
@@ -70,16 +83,7 @@ def convert_file(filepath):
 
     print(repr(post.metadata))
 
-    # Handle highlights
-    body = post.content
-    body = re.sub(r'\{% highlight( (\S+)) %\}', r'```\2', body)
-    body = re.sub(r'\{% endhighlight %\}', r'```', body)
-    post.content = body
-
-    # Handle assets?
-
-
-    # Generate new filepath and write
+    # Generate new filepath
     pathparts = filepath.split("/")
     pathparts[0] = GRAVDIR
     pathparts[1] = FIRST_LEVEL[pathparts[1]]
@@ -91,9 +95,29 @@ def convert_file(filepath):
     newfilepath = "/".join(pathparts)
     print("Creating {}...".format(newfilefolder))
     os.makedirs(newfilefolder, exist_ok=True)
+    target_path = newfilefolder
+
+    # BODY HANDLING
+    body = post.content
+
+    # Handle highlights
+    body = re.sub(r'\{% highlight( (\S+)) %\}', r'```\2', body)
+    body = re.sub(r'\{% endhighlight %\}', r'```', body)
+
+    # Handle assets?
+    # Images OLD: ![alt text]({{ site.url }}/assets/blah.jpg "title")
+    # Images NEW: ![alt text](blah.jpg "title")
+    body = re.sub(r'!\[(.*?)\]\((.+?)( [\'"].+[\'"])?\)', replace_image, body)
+
+    # FINISH BODY HANDLING
+    post.content = body
+
+    # Write to new location
     print("Writing {}...".format(newfilepath))
     frontmatter.dump(post, newfilepath)
 
+
+# MAIN SCRIPT
 num_converted = 0
 for root, dirs, files in os.walk(SRCDIR):
     if root.split("/")[-1] in ["assets", "css", "images", "fonts", "javascripts", "_includes", "_layouts", SRCDIR]:
@@ -106,4 +130,3 @@ for root, dirs, files in os.walk(SRCDIR):
         num_converted += 1
 
 print("Converted {} files.".format(num_converted))
-