a text output (and skel an outline)HEADmain

- spine --text [--output=output path] [markup source]

1 files changed, 18 insertions, 16 deletions

diff --git a/org/default_regex.org b/org/default_regex.org
index 9a0ded7..fb2b993 100644
--- a/org/default_regex.org
+++ b/org/default_regex.org
@@ -308,9 +308,9 @@ static table_col_separator_nl                   = ctRegex!(`[┊]$`, "mg");
 #+BEGIN_SRC d
 /+ inline markup footnotes endnotes +/
 static inline_notes_curly_gen                   = ctRegex!(`~\{.+?\}~`, "m");
-static inline_notes_curly                       = ctRegex!(`~\{\s*(.+?)\}~`, "mg");
-static inline_notes_curly_sp_asterisk           = ctRegex!(`~\{[*]+\s+(.+?)\}~`, "m");
-static inline_notes_curly_sp_plus               = ctRegex!(`~\{[+]+\s+(.+?)\}~`, "m");
+static inline_notes_curly                       = ctRegex!(`~\{\s*(.+?)\s*\}~`, "mg");
+static inline_notes_curly_sp_asterisk           = ctRegex!(`~\{[*]+\s+(.+?)\s*\}~`, "m");
+static inline_notes_curly_sp_plus               = ctRegex!(`~\{[+]+\s+(.+?)\s*\}~`, "m");
 static note_ref                                 = ctRegex!(`^\S+?noteref_(?P<ref>[0-9]+)`, "mg");     // {^{73.}^}#noteref_73
 #+END_SRC
 
@@ -331,8 +331,8 @@ static smid_inline_url_generic                        = ctRegex!(`(?:^|[}(\[ ])(
 static smid_inline_url                                = ctRegex!(`((?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)[a-zA-Z0-9_]\S*)`, "mg");
 static smid_inline_link_naked_url                     = ctRegex!(`(?P<pre>^|[ (\[])(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤)\S+?)(?=[.,;:?!'"]?([ )\]]|$))`, "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");
+static smid_inline_link_endnote_url_helper_punctuated = ctRegex!(`\{~\^\s+(?P<content>.+?)\s*\}(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+?)(?=[.,;:?!]?([ ]|$))`, "mg");
+static smid_inline_link_endnote_url_helper            = ctRegex!(`\{~\^\s+(?P<content>.+?)\s*\}(?P<link>(?:(?:https?|git):\/\/|¤?\.\.\/|¤?\.\/|¤|#)\S+)`, "mg");
 #+END_SRC
 
 #+BEGIN_SRC d
@@ -507,7 +507,6 @@ static make_breakcolumn                         = ctRegex!(`break=(?P<breakcolum
 #+END_SRC
 
 * 2. ctRegex defaults shared by meta & output (generic)
-
 ** meta
 
 #+NAME: prgmkup_rgx_meta
@@ -545,9 +544,9 @@ static middle_dot                               = ctRegex!(`·`, "mg");
 static br_empty_line                            = ctRegex!(`\n[ ]*\n`, "mg");
 static br_linebreaks_newlines                   = ctRegex!(`[\n┘┙]`, "mg");
 static br_linebreaks                            = ctRegex!(`[┘┙]`, "mg");
-static br_line                                  = ctRegex!(`┘`, "mg");
-static br_line_inline                           = ctRegex!(`┙`, "mg");
-static br_line_spaced                           = ctRegex!(`┚`, "mg");
+static br_line                                  = ctRegex!(`\s*┘\s*`, "mg");
+static br_line_inline                           = ctRegex!(`\s*┙\s*`, "mg");
+static br_line_spaced                           = ctRegex!(`\s*┚\s*`, "mg");
 #+END_SRC
 
 #+BEGIN_SRC d
@@ -604,6 +603,8 @@ static inline_notes_al_gen                      = ctRegex!(`【.+?】`, "m");
 static inline_notes_al_gen_text                 = ctRegex!(`【(?P<text>.+?)】`, "m");
 static inline_notes_al_all_note                 = ctRegex!(`【(?P<num>\d+|(?:[*]|[+])+)\s+(?P<note>.+?)\s*】`, "mg");
 static inline_notes_al_regular_number_note      = ctRegex!(`【(?P<num>\d+)\s+(?P<note>.+?)\s*】`, "mg");
+// static inline_notes_al_all_note                 = ctRegex!(`【(?P<num>\d+|(?:[*]|[+])+)\s+(?P<note>.+?)\s*(≫\s\d+)?\s*】`, "mg"); // ocn of origin would be useful in endnote section
+// static inline_notes_al_regular_number_note      = ctRegex!(`【(?P<num>\d+)\s+(?P<note>.+?)\s*(≫\s\d+)?\s*】`, "mg"); // ocn of origin would be useful in endnote section
 static inline_notes_al_special_char_note        = ctRegex!(`【(?P<char>(?:[*]|[+])+)\s+(?P<note>.+?)】`, "mg");
 static inline_al_delimiter_open_regular         = ctRegex!(`【\s`, "m");
 static inline_al_delimiter_open_symbol_star     = ctRegex!(`【[*]\s`, "m");
@@ -625,13 +626,14 @@ static inline_image                             = ctRegex!(`(?P<pre>┥)☼(?P<i
 static inline_image_without_dimensions          = ctRegex!(`(?P<pre>┥)☼(?P<imginf>(?P<img>[a-zA-Z0-9._-]+?\.(?:jpg|gif|png)),w(?P<width>0)h(?P<height>0))\s*(?P<post>.*?┝┤.*?├)`, "mg");
 static inline_image_info                        = ctRegex!(`☼?(?P<img>[a-zA-Z0-9._-]+?\.(?:jpg|gif|png)),w(?P<width>\d+)h(?P<height>\d+)`, "mg");
 static inline_link_anchor                       = ctRegex!(`┃(?P<anchor>\S+?)┃`, "mg"); // TODO *~text_link_anchor
-static inline_link                              = ctRegex!(`┥(?P<text>.+?)┝┤(?P<link>#?(\S+?))├`, "mg");
-static inline_link_empty                        = ctRegex!(`┥(?P<text>.+?)┝┤├`, "mg");
-static inline_link_number                       = ctRegex!(`┥(?P<text>.+?)┝┤(?P<num>[0-9]+)├`, "mg"); // not used
-static inline_link_number_only                  = ctRegex!(`(?P<linked_text>┥.+?┝)┤(?P<num>[0-9]+)├`, "mg");
-static inline_link_stow_uri                     = ctRegex!(`┥(?P<text>.+?)┝┤(?P<link>[^ 0-9#┥┝┤├][^ 0-9┥┝┤├]+)├`, "mg"); // will not stow (stowed links) or object number internal links
-static inline_link_hash                         = ctRegex!(`┥(?P<text>.+?)┝┤(?P<link>#(?P<hash>\S+?))├`, "mg");
-static inline_link_seg_and_hash                 = ctRegex!(`┥(?P<text>.+?)┝┤(?P<link>(?P<seg>[^/#├]*)#(?P<hash>.+?))├`, "mg");
+// space cleaning should not be necessary
+static inline_link                              = ctRegex!(`┥\s*(?P<text>.+?)\s*┝┤(?P<link>#?(\S+?))├`, "mg");
+static inline_link_empty                        = ctRegex!(`┥\s*(?P<text>.+?)\s*┝┤├`, "mg");
+static inline_link_number                       = ctRegex!(`┥\s*(?P<text>.+?)\s*┝┤(?P<num>[0-9]+)├`, "mg"); // not used
+static inline_link_number_only                  = ctRegex!(`\s*(?P<linked_text>\s*┥.+?┝)┤(?P<num>[0-9]+)├`, "mg");
+static inline_link_stow_uri                     = ctRegex!(`┥\s*(?P<text>.+?)\s*┝┤(?P<link>[^ 0-9#┥┝┤├][^ 0-9┥┝┤├]+)├`, "mg"); // will not stow (stowed links) or object number internal links
+static inline_link_hash                         = ctRegex!(`┥\s*(?P<text>.+?)\s*┝┤(?P<link>#(?P<hash>\S+?))├`, "mg");
+static inline_link_seg_and_hash                 = ctRegex!(`┥\s*(?P<text>.+?)\s*┝┤(?P<link>(?P<seg>[^/#├]*)#(?P<hash>.+?))├`, "mg");
 static inline_link_clean                        = ctRegex!(`┤(?:.+?)├|[┥┝]`, "mg");
 static inline_link_toc_to_backmatter            = ctRegex!(`┤#(?P<link>endnotes|bibliography|bookindex|glossary|blurb)├`, "mg");
 static url                                      = ctRegex!(`https?://`, "mg");