support for root links, various fixes on relative links, added doku shortocde

layouts/shortcodes/doku.html

@@ -0,0 +1 @@
+<a href="dokuwiki.org/wiki:{{ index .Params 0 }}" target="_blank">{{ index .Params 0 }}</a>

src/dokuwiki_to_hugo.py

@@ -6,9 +6,11 @@ from src.markdown_converter import MarkdownConverter
 
 
 class DokuWikiToHugo:
+    root_dir = ""
 
-    def __init__(self):
+    def __init__(self, root = None):
         self.header_converter = HugoFileConfig()
+        DokuWikiToHugo.root_dir = root
         pass
 
     def create_output_dir(self):
@@ -20,6 +22,7 @@ class DokuWikiToHugo:
     def doku_to_hugo(self, dir):
         self.create_output_dir()
         for root, subFolders, files in os.walk(dir):
+            files = [f for f in files if not f[0] == '.']
             for file in files:
                 self.process_file(root, file)
 

src/main.py

@@ -3,11 +3,12 @@ import argparse
 from src.dokuwiki_to_hugo import DokuWikiToHugo
 
 
-def main(dir):
-    DokuWikiToHugo().doku_to_hugo(dir)
+def main(dir, root):
+    DokuWikiToHugo(root).doku_to_hugo(dir)
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser()
-    parser.add_argument("-d", "--dir", help="the root directory of your DokuWiki pages collection", required = True)
+    parser.add_argument("-d", "--dir", help="the directory of your DokuWiki pages collection", required = True)
+    parser.add_argument("-r", "--root", help="the root directory name in your hugo content where to link to", required = False)
     opts = parser.parse_args()
-    main(opts.dir)
+    main(opts.dir, opts.root)

src/markdown/headers.py

@@ -19,6 +19,8 @@ class MarkdownHeader():
         result = text
         for regex_head in MarkdownHeader.pattern.findall(text):
             orig_header = ''.join(regex_head)
-            new_header = ('#' * MarkdownHeader.config[regex_head[0]]) + regex_head[1]
-            result = result.replace(orig_header, new_header)
+            src_header = regex_head[0]
+            if src_header in MarkdownHeader.config:
+                new_header = ('#' * MarkdownHeader.config[src_header]) + regex_head[1]
+                result = result.replace(orig_header, new_header)
         return result

src/markdown/links.py

@@ -2,6 +2,7 @@ import re
 from os import walk
 from pathlib import Path
 
+from src.dokuwiki_to_hugo import DokuWikiToHugo
 from src.markdown_converter import MarkdownConverter
 
 
@@ -12,7 +13,13 @@ class MarkdownLinks():
 
     def convert(self, text):
         result = text
+        def starts_with_space(match):
+            return match[1][0] is ' '
+
         for regex_link in MarkdownLinks.pattern.findall(text):
+            if starts_with_space(regex_link):
+                continue
+
             origlink = ''.join(regex_link)
             convertedlink = ""
             if "http" in origlink or "www" in origlink:
@@ -27,8 +34,17 @@ class MarkdownLinks():
     def parseUrl(self, text):
         return text[2:text.index('|')]
 
+    def add_md_and_replace_home_with_index(self, src_url):
+        url = src_url
+        if "." not in url:
+            url = url + ".md"
+        return url.replace('home.md', '_index.md')
     def parseInternalUrl(self, text):
-        return text[2:len(text)-2].replace(":", "/")
+        url = text[2:len(text) - 2].replace(":", "/")
+        return self.add_md_and_replace_home_with_index(url)
+    def parseInternalUrlWithoutTitle(self, text):
+        url = self.parseUrl(text).replace(":", "/")
+        return self.add_md_and_replace_home_with_index(url)
 
     def parseTitle(self, text):
         return text[text.index('|') + 1: text.index(']]')]
@@ -38,16 +54,24 @@ class MarkdownLinks():
         self.assert_interwiki_is_known(interwiki_shortcode)
         interwiki_urlpart = text[text.index('>') + 1 : len(text) - 2]
 
-        return """{{< %s %s >}}""" % (interwiki_shortcode, interwiki_urlpart)
+        return """{{< %s "%s" >}}""" % (interwiki_shortcode, interwiki_urlpart)
+
+    def root_dir(self, url):
+        if "/" not in url or DokuWikiToHugo.root_dir is None or url[0] is "/":
+            return ""
+        return DokuWikiToHugo.root_dir + "/"
 
     def convert_as_internal_link(self, text):
+        url = ""
+        title = ""
         if "|" not in text:
-            return """{{< relref "%s" >}}""" % (self.parseInternalUrl(text))
-
-        url = self.parseUrl(text).replace(":", "/")
-        title = self.parseTitle(text)
+            url = self.parseInternalUrl(text)
+            title = text[2:len(text)-2].replace(":", "/")
+        else:
+            url = self.parseInternalUrlWithoutTitle(text)
+            title = self.parseTitle(text)
 
-        return """[%s]({{< relref "%s" >}})""" % (title, url)
+        return """[%s]({{< relref "%s%s" >}})""" % (title, self.root_dir(url), url.replace(' ', '_'))
 
     def convert_as_external_link(self, text):
         if '|' in text:

test/markdown/test_header.py

@@ -8,6 +8,10 @@ class TestMarkdownHeader(TestCase):
     def setUp(self):
         self.converter = MarkdownHeader()
 
+    def test_does_not_convert_if_too_long(self):
+        src = "# ======================================="
+        self.assertEqual(src, self.converter.convert(src))
+
     def test_convert_does_nothing_if_no_header(self):
         self.assertEqual("blabla", self.converter.convert("blabla"))
 

test/markdown/test_links.py

@@ -1,5 +1,6 @@
 from unittest import TestCase
 
+from src.dokuwiki_to_hugo import DokuWikiToHugo
 from src.markdown.links import MarkdownLinks
 
 
@@ -8,62 +9,110 @@ class TestMarkdownLinks(TestCase):
     def setUp(self):
         self.converter = MarkdownLinks()
 
+    def tearDown(self):
+        DokuWikiToHugo.root_dir = None
+
     def text_convert_no_link_returns_text(self):
         self.assertEquals("sup", self.converter.convert("sup"))
 
-    def test_internal_links_without_text_converted_properly(self):
-        mdLink = """{{< relref "bla" >}}"""
-        dokuLink = "[[bla]]"
+    def test_do_not_convert_starting_with_space(self):
+        # I actually should be looking for ``` to see if this shouldn't be converted
+        # but that would be way too complex for a simple onetime conversion so fuck that.
+        no_real_doku_link = """if [[ "$line" =~ "|" ]]; then"""
+        self.assertEqual(no_real_doku_link, self.converter.convert(no_real_doku_link))
+
+    def test_internal_link_with_space_converted_to_underscore(self):
+        mdLink = """[magic properties]({{< relref "magic_properties.md" >}})"""
+        dokuLink = "[[magic properties]]"
 
         self.assertEqual(mdLink, self.converter.convert(dokuLink))
 
-    def test_unknown_interwiki_link_throws_exception(self):
-        with self.assertRaises(ValueError):
-            self.converter.convert("[[whaddapdawg>Wiki]]")
+    def test_internal_link_to_home_converted_to_index(self):
+        mdLink = """[subpage/home]({{< relref "subpage/_index.md" >}})"""
+        dokuLink = "[[subpage/home]]"
 
-    def test_known_interwiki_link_with_some_spaces(self):
-        mdLink = """{{< lib Purple Cow: Transform Your Business by Being Remarkable >}}"""
-        dokuLink = "[[lib>Purple Cow: Transform Your Business by Being Remarkable]]"
+        self.assertEqual(mdLink, self.converter.convert(dokuLink))
+
+    def test_internal_link_with_ref_to_root_not_prefixed_with_root_dir(self):
+        DokuWikiToHugo.root_dir = "pages"
+        mdLink = """[Download SaveGame #one]({{< relref "/games/Wizardry8/Wizardry8_Saves01.rar" >}})"""
+        dokuLink = """[[/games/Wizardry8/Wizardry8_Saves01.rar|Download SaveGame #one]]"""
 
         self.assertEqual(mdLink, self.converter.convert(dokuLink))
 
-    def test_known_interwiki_link_converts_successfully(self):
-        # see https://gohugo.io/extras/shortcodes/
-        mdLink = """{{< wp Wiki >}}"""
-        dokuLink = "[[wp>Wiki]]"
+    def test_internal_link_prefixed_with_configured_root_dir_if_subdir(self):
+        DokuWikiToHugo.root_dir = "pages"
+        mdLink = """[bla/blie]({{< relref "pages/bla/blie.md" >}})"""
+        dokuLink = "[[bla:blie]]"
+
+        self.assertEqual(mdLink, self.converter.convert(dokuLink))
+
+    def test_internal_links_not_prefixed_with_configured_root_dir_if_currdir(self):
+        DokuWikiToHugo.root_dir = "pages"
+        mdLink = """[bla]({{< relref "bla.md" >}})"""
+        dokuLink = "[[bla]]"
+
+        self.assertEqual(mdLink, self.converter.convert(dokuLink))
+
+    def test_internal_links_without_text_converted_properly(self):
+        mdLink = """[bla]({{< relref "bla.md" >}})"""
+        dokuLink = "[[bla]]"
 
         self.assertEqual(mdLink, self.converter.convert(dokuLink))
 
     def test_internal_links_with_sublink_converted_properly(self):
-        mdLink = """[text]({{< relref "bla/blie" >}})"""
+        mdLink = """[text]({{< relref "bla/blie.md" >}})"""
         dokuLink = "[[bla:blie|text]]"
 
         self.assertEqual(mdLink, self.converter.convert(dokuLink))
 
     def test_multiple_links_in_text_converted_properly(self):
-        mdLink = """{{< relref "bla" >}} wow this looks cool and so does {{< relref "this" >}} and such"""
+        mdLink = """[bla]({{< relref "bla.md" >}}) wow this looks cool and so does [this]({{< relref "this.md" >}}) and such"""
         dokuLink = "[[bla]] wow this looks cool and so does [[this]] and such"
 
         self.assertEqual(mdLink, self.converter.convert(dokuLink))
 
     def test_internal_links_with_some_text_in_line_converted_properly(self):
-        mdLink = """[text]({{< relref "bla" >}}) wow this looks cool"""
+        mdLink = """[text]({{< relref "bla.md" >}}) wow this looks cool"""
         dokuLink = "[[bla|text]] wow this looks cool"
 
         self.assertEqual(mdLink, self.converter.convert(dokuLink))
 
     def test_internal_link_with_image_tag(self):
         dokuLink = "[[code|<img src='code.jpg'>]]"
-        mdLink = """[<img src='code.jpg'>]({{< relref "code" >}})"""
+        mdLink = """[<img src='code.jpg'>]({{< relref "code.md" >}})"""
+
+        self.assertEqual(mdLink, self.converter.convert(dokuLink))
+
+    def test_internal_links_with_extension_not_suffixed_with_md(self):
+        mdLink = """[text]({{< relref "bla.zip" >}})"""
+        dokuLink = "[[bla.zip|text]]"
 
         self.assertEqual(mdLink, self.converter.convert(dokuLink))
 
     def test_internal_links_converted_properly(self):
-        mdLink = """[text]({{< relref "bla" >}})"""
+        mdLink = """[text]({{< relref "bla.md" >}})"""
         dokuLink = "[[bla|text]]"
 
         self.assertEqual(mdLink, self.converter.convert(dokuLink))
 
+    def test_unknown_interwiki_link_throws_exception(self):
+        with self.assertRaises(ValueError):
+            self.converter.convert("[[whaddapdawg>Wiki]]")
+
+    def test_known_interwiki_link_with_some_spaces(self):
+        mdLink = """{{< lib "Purple Cow: Transform Your Business by Being Remarkable" >}}"""
+        dokuLink = "[[lib>Purple Cow: Transform Your Business by Being Remarkable]]"
+
+        self.assertEqual(mdLink, self.converter.convert(dokuLink))
+
+    def test_known_interwiki_link_converts_successfully(self):
+        # see https://gohugo.io/extras/shortcodes/
+        mdLink = """{{< wp "Wiki" >}}"""
+        dokuLink = "[[wp>Wiki]]"
+
+        self.assertEqual(mdLink, self.converter.convert(dokuLink))
+
     def test_external_link_without_title(self):
         mdLink = "[https://www.google.com](https://www.google.com)"
         dokuLink = "[[https://www.google.com]]"