regex, resolve issues with: master file inserts path; url internal link

1 files changed, 15 insertions, 15 deletions

diff --git a/org/default_regex.org b/org/default_regex.org
index 7f0c724..1fd728a 100644
--- a/org/default_regex.org
+++ b/org/default_regex.org
@@ -146,8 +146,8 @@ static heading_seg_and_above                          = ctRegex!(`^:?([A-D1])[~]
 static heading_marker                                 = ctRegex!(`^:?([A-D1-4])[~]`);
 static heading_anchor_tag                             = ctRegex!(`^:?[A-D1-4][~]([a-z0-9_.-]+) `,"i");
 static heading_identify_anchor_tag                    = ctRegex!(`^:?[A-D1-4][~]\s+(?:(?:(?:chapter|article|section|clause)\s+[0-9.]+)|(?:[0-9]+))`,"i");
-static heading_extract_named_anchor_tag               = ctRegex!(`^:?[A-D1-4][~]\s+(chapter|article|section|clause)\s+((?:[0-9]+[.:])*[0-9]+)(?:[.:;, ]|$)`,"i");
-static heading_extract_unnamed_anchor_tag             = ctRegex!(`^:?[A-D1-4][~]\s+((?:[0-9]+.)*[0-9]+)(?:[.:;, ]|$)`);
+static heading_extract_named_anchor_tag               = ctRegex!(`^:?[A-D1-4][~]\s+(chapter|article|section|clause)\s+((?:[0-9]+[.:])*[0-9]+)(?=[.:;, ]|$)`,"i");
+static heading_extract_unnamed_anchor_tag             = ctRegex!(`^:?[A-D1-4][~]\s+((?:[0-9]+.)*[0-9]+)(?=[.:;, ]|$)`);
 static heading_marker_missing_tag                     = ctRegex!(`^:?([A-D1-4])[~] `);
 static heading_anchor_tag_plus_colon                  = ctRegex!(`^:?([A-D1-4][~])([a-z0-9_.:-]+) `,"i");
 static heading_marker_tag_has_colon                   = ctRegex!(`([:])`);
@@ -168,7 +168,7 @@ static para_attribs                                   = ctRegex!(`^_(?:(?:[0-9])
 #+END_SRC
 
 ** blocked markup
-*** blocked markup                                              :block:tic:
+*** blocked markup curly & tic                                      :block:
 
 #+name: meta_rgx
 #+BEGIN_SRC d
@@ -177,7 +177,7 @@ static block_open                                     = ctRegex!("^((code([.][a-
 static block_poem_open                                = ctRegex!("^((poem[{].*?$)|`{3} poem)");
 #+END_SRC
 
-** blocked markup tics                                           :block:tic:
+*** blocked markup tic                                          :block:tic:
 
 #+name: meta_rgx
 #+BEGIN_SRC d
@@ -260,9 +260,9 @@ static note_ref                                       = ctRegex!(`^\S+?noteref_(
 static webserv_url_doc_root                           = ctRegex!(`(?P<url>(?P<domain>https?:\/\/[^ /]+)\/(?P<path>\S*))`, "mg");
 static smid_inline_url_generic                        = ctRegex!(`(?:^|[}(\[ ])(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)[a-zA-Z0-9_#]`, "mg");
 static smid_inline_url                                = ctRegex!(`((?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)[a-zA-Z0-9_]\S*)`, "mg");
-static smid_inline_link_naked_url                     = ctRegex!(`(?P<pre>^|[ ])(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤)\S+?)(?P<post>[.,;:?!'"]?(?:[ ]|$))`, "mg"); // issue with #link #32
-static smid_inline_link_markup_regular                = ctRegex!(`(?P<pre>^|[ ]|[^\S]?)\{\s*(?P<content>.+?)\s*\}(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?P<post>[;:!,?.]?(?:[ )\]]|$))`, "mg"); // NEXT
-static smid_inline_link_endnote_url_helper_punctuated = ctRegex!(`\{~\^\s+(?P<content>.+?)\}(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?P<after>[.,;:?!]?(?:[ ]|$))`, "mg");
+static smid_inline_link_naked_url                     = ctRegex!(`(?P<pre>^|[ (\[])(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤)\S+?)(?=[.,;:?!'"]?([ )\]]|$))`, "mg");
+static smid_inline_link_markup_regular                = ctRegex!(`(?P<pre>^|[ (\[])\{\s*(?P<content>.+?)\s*\}(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?=[;:!,?.]?([ )\]]|$))`, "mg");
+static smid_inline_link_endnote_url_helper_punctuated = ctRegex!(`\{~\^\s+(?P<content>.+?)\}(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?=[.,;:?!]?([ ]|$))`, "mg");
 static smid_inline_link_endnote_url_helper            = ctRegex!(`\{~\^\s+(?P<content>.+?)\}(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+)`, "mg");
 #+END_SRC
 
@@ -271,10 +271,10 @@ static smid_inline_link_endnote_url_helper            = ctRegex!(`\{~\^\s+(?P<co
 #+name: meta_rgx
 #+BEGIN_SRC d
 static image                                           = ctRegex!(`([a-zA-Z0-9._-]+?\.(?:png|gif|jpg))`, "mg");
-static smid_image_generic                              = ctRegex!(`(?:^|[ ]|[^\S]?)\{(?:~\^\s+|\s*)\S+\.(?:png|gif|jpg).+?\}(?:image|(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)[;:!,?.]?(?:[ )\]]|$)`, "mg");
-static smid_image_with_dimensions                      = ctRegex!(`(?P<pre>(?:^|[ ]|[^\S]?)\{(?:~\^\s+|\s*))(?P<image>\S+\.(?:png|gif|jpg))\s+(?P<width>\d+)x(?P<height>\d+)\s*(?P<post>(?:.+?)\s*\}(?:image|(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?:[;:!,?.]?(?:[ )\]]|$)))`, "mg");
-static smid_image                                      = ctRegex!(`(?P<pre>(?:^|[ ]|[^\S]?)\{(?:~\^\s+|\s*))(?P<image>\S+\.(?:png|gif|jpg))\s*(?P<post>(?:.+?)\s*\}(?:image|(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?:[;:!,?.]?(?:[ )\]]|$)))`, "mg");
-static smid_mod_image_without_dimensions               = ctRegex!(`\{(?:~\^\s+|\s*)☼\S+\.(?:png|gif|jpg),w0h0\s+(?:.+?)\s*\}(?:image|(?:https?|git):\/\/\S+?)(?:[;:!,?.]?(?:[ )\]]|$))`, "mg");
+static smid_image_generic                              = ctRegex!(`(?:^|[ ]|[^\S]?)\{(?:~\^\s+|\s*)\S+\.(?:png|gif|jpg).+?\}(?:image|(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?=[;:!,?.]?([ )\]]|$))`, "mg");
+static smid_image_with_dimensions                      = ctRegex!(`(?P<pre>(?:^|[ ]|[^\S]?)\{(?:~\^\s+|\s*))(?P<image>\S+\.(?:png|gif|jpg))\s+(?P<width>\d+)x(?P<height>\d+)\s*(?P<post>(?:.+?)\s*\}(?:image|(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?=[;:!,?.]?([ )\]]|$)))`, "mg");
+static smid_image                                      = ctRegex!(`(?P<pre>(?:^|[ ]|[^\S]?)\{(?:~\^\s+|\s*))(?P<image>\S+\.(?:png|gif|jpg))\s*(?P<post>(?:.+?)\s*\}(?:image|(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?=[;:!,?.]?([ )\]]|$)))`, "mg");
+static smid_mod_image_without_dimensions               = ctRegex!(`\{(?:~\^\s+|\s*)☼\S+\.(?:png|gif|jpg),w0h0\s+(?:.+?)\s*\}(?:image|(?:https?|git):\/\/\S+?)(?=[;:!,?.]?([ )\]]|$))`, "mg");
 #+END_SRC
 
 *** inline markup book index                             :inline:bookindex:
@@ -439,10 +439,10 @@ static pod_content_location                           =
   ctRegex!(`^(?P<filename>[a-zA-Z0-9._-]+[.]ss[tm])(?P<languages>(?:\s+[a-z]{2}(?:,|$))+)`, "mg");
 static src_fn                                         =
   ctRegex!(`^([/]?(?:[a-zA-Z0-9._-]+/)*)(?P<fn_src>(?P<fn_base>[a-zA-Z0-9._-]+)[.](?P<fn_src_suffix>ss[tm]))$`);
-static src_fn_master                                  = ctRegex!(`^(?P<path>[a-zA-Z0-9._-]+/)*(?P<filename>[a-zA-Z0-9._-]+[.]ssm)$`);
-static src_fn_text                                    = ctRegex!(`^(?P<path>[a-zA-Z0-9._-]+/)*(?P<filename>[a-zA-Z0-9._-]+[.]sst)$`);
-static src_fn_insert                                  = ctRegex!(`^(?P<path>[a-zA-Z0-9._-]+/)*(?P<filename>[a-zA-Z0-9._-]+[.]ssi)$`);
-static src_fn_find_inserts                            = ctRegex!(`^(?P<path>[a-zA-Z0-9._-]+/)*(?P<filename>[a-zA-Z0-9._-]+[.]ss[im])$`);
+static src_fn_master                                  = ctRegex!(`^(?P<path>/?(?:[a-zA-Z0-9._-]+/)*)(?P<filename>[a-zA-Z0-9._-]+[.]ssm)$`);
+static src_fn_text                                    = ctRegex!(`^(?P<path>/?(?:[a-zA-Z0-9._-]+/)*)(?P<filename>[a-zA-Z0-9._-]+[.]sst)$`);
+static src_fn_insert                                  = ctRegex!(`^(?P<path>/?(?:[a-zA-Z0-9._-]+/)*)(?P<filename>[a-zA-Z0-9._-]+[.]ssi)$`);
+static src_fn_find_inserts                            = ctRegex!(`^(?P<path>/?(?:[a-zA-Z0-9._-]+/)*)(?P<filename>[a-zA-Z0-9._-]+[.]ss[im])$`);
 static insert_src_fn_ssi_or_sst                       = ctRegex!(`^<<\s*(?P<path>[a-zA-Z0-9._-]+/)*(?P<filename>[a-zA-Z0-9._-]+[.]ss[ti])$`);
 static src_base_parent_dir_name                       = ctRegex!(`[/](?P<dir>(?:[a-zA-Z0-9._-]+))(?:/media/text/[a-z]{2})$`); // formalizes dir structure
 static src_base_parent_path                           = ctRegex!(`(?P<dir>(?:[/a-zA-Z0-9._-]+))(?:/media/text/[a-z]{2})$`); // formalizes dir structure